Cognira

cognira is an independent AI assistant built around user-taught memory, which may appeal to people who want to see what an assistant remembers and control that memory directly.

Quick Verdict

cognira lets users teach information in a sentence, then uses that information from the next reply onward. Replies that use a memory identify it, and users can forget a memory with one tap. The hosted assistant is free, but privacy-conscious users should review how conversation content may be used and verify retention, deletion, and security details.

The project is independently served from one 16GB iMac, according to the website. That is a concrete trust and transparency signal, but it also raises practical questions about capacity, availability, and performance that the website does not clearly answer.

Key Features

• Teach in plain language: Users can provide information in a sentence rather than managing a separate knowledge base.
• Immediate use: Information taught to cognira is used from the next reply.
• Visible memory use: Replies identify the memory they used, helping users understand why an answer may reflect earlier teaching.
• One-tap forgetting: Users can remove a memory with one tap.
• Local option: cognira Entity runs the same models locally on the user's machine, offering an alternative to the hosted assistant.
• Free hosted assistant: The website describes the hosted cognira assistant as free.

The website describes cognira as self-learning and low-energy, but it does not clearly state the models' capabilities, performance, technical specifications, or availability requirements for cognira Entity. Those details should be verified before relying on it for demanding work.

Who this is for

cognira appears best suited to curious individuals who want to teach an AI assistant personal information, inspect when that information affects replies, and remove it without a complex settings process. It may also interest users who prefer an independently run project or want to investigate a local option.

It is a less obvious fit for organizations requiring documented security controls, defined retention and deletion policies, predictable capacity, or detailed technical specifications. Eligible conversation content may help train a future model version, while the privacy policy does not specify which conversation parts qualify. Verify those terms on the official website before sharing sensitive information.